Looking Ahead

The near-term trajectory of agentic AI is visible in the deployments that exist today. The longer-term trajectory requires reading the direction those deployments are pointing.

By 2030, Gartner’s analysis suggests that AI agents will be as commonplace in the enterprise as personal computers. That analogy is more precise than it first appears. The personal computer did not simply make individual workers more productive — it transformed the organizational structures, job functions, roles, and business models that existed around it. Agentic AI is on a similar trajectory: not just making existing processes faster, but making fundamentally different organizational configurations possible.

New Roles and Organizational Structures

The emergence of agentic AI is already generating new professional roles that did not exist three years ago. These roles are not uniformly technical — they span operational, managerial, and strategic functions.

Agent operations specialists manage the monitoring, performance evaluation, and escalation handling for production agentic systems. Like network operations centers that existed before them, these functions ensure that agent fleets operating at scale are performing within defined parameters and that anomalies are detected and addressed promptly.

Agent designers and prompt engineers have matured from experimental roles into structured engineering disciplines, with organizations developing internal standards for how agents are instructed, constrained, and evaluated before deployment.

AI ethics officers and governance leads have moved from advisory functions to line roles with authority over deployment approvals. In regulated industries, these positions are increasingly tied to compliance accountability — not just advisory influence.

Human-agent workflow architects design the handoff points between human professionals and autonomous agents, specifying the conditions under which agents escalate, the information that must be present for agents to proceed, and the interfaces through which humans review agent outputs and provide direction. This is a design discipline that draws on organizational behavior, process engineering, and AI systems knowledge simultaneously.

Organizations that treat these roles as temporary or transitional are misreading the trend. The blended workforce — humans and agents operating in structured collaboration — is not a transitional state on the way to full automation. It is the durable organizational model for enterprises that want the efficiency of agentic AI while maintaining the judgment, accountability, and adaptability that human professionals provide.

The Regulatory Landscape Is Taking Shape

In 2025, the regulatory environment for agentic AI moved from vague guidance to specific requirements in several jurisdictions. The EU AI Act’s risk-based framework for high-stakes AI systems has begun to affect procurement and deployment decisions globally. OMB’s memos in the United States established specific requirements for government AI use — including model training prohibitions, Chief AI Officer authority, and human-in-the-loop mandates for citizen-affecting decisions.

The direction of travel is clear: regulatory frameworks for agentic AI will become more specific, more enforceable, and more consequential for organizations that have not built compliance capacity in advance. The organizations that are actively engaging with regulators, participating in standards bodies, and building governance frameworks that exceed current requirements will be best positioned when those requirements become mandatory.

Cross-industry collaboration on agentic AI standards is also accelerating. Healthcare consortia are developing shared evaluation frameworks for clinical AI performance. Government technology bodies are publishing shared architectures for FedRAMP-compliant agentic deployments. Financial regulators are developing explainability requirements for algorithmic decision-making that will extend to agentic systems. Organizations that engage with these collaborative standard-setting processes help shape frameworks that reflect operational realities, rather than inheriting requirements they had no role in defining.

Multi-Modal Agents and Expanded Capability

The agentic systems described throughout this playbook operate primarily through language: reading documents, generating text, executing tool calls, and coordinating through structured communication. The next generation of agentic systems integrates multiple modalities — vision, audio, structured data, and code execution — in ways that expand what agents can perceive and act upon.

Multi-modal agents can analyze medical imaging alongside clinical text, review product defects from visual inspection feeds, process spoken customer service interactions without transcription intermediaries, and operate software interfaces designed for humans rather than APIs. These capabilities are emerging in production deployments in 2025 and will become broadly available over the following years.

The implications for enterprise architecture are significant. Processes that previously required human involvement because they involved visual inspection, audio interpretation, or interaction with interfaces that lacked APIs are candidates for agentic augmentation as multi-modal capability matures. Organizations that have built the governance infrastructure for language-based agentic AI will find it transferable to multi-modal systems — the principles are the same, even as the perceptual and action capabilities expand.

The Long View

The vision for agentic AI in 2030 is not a world from which human judgment has been removed. It is a world in which human judgment is applied more selectively, more effectively, and at higher levels of consequence than is currently possible. The routine, the repetitive, the data-intensive, and the coordination-heavy functions of enterprise work migrate to agents. The complex, the novel, the relational, and the accountable functions remain with human professionals — augmented by agents that give them better information, faster synthesis, and greater operational reach.

Organizations that are building toward this future now — with deliberate strategy, sound governance, and authentic investment in their workforces’ capacity to lead alongside AI — are not just improving their operational efficiency. They are building the institutional capability that will define competitive standing in their sectors for the decade ahead.

Make It Your Own

Key questions to ask in the context of your organization:

  • Have you identified the new roles that agentic AI will create in your organization — agent operations, workflow architecture, governance oversight — and begun developing hiring and training plans to fill them?
  • Is your organization actively monitoring the regulatory trajectory for agentic AI in your sector, and do you have a plan to build compliance capacity before new requirements become mandatory?
  • Are you participating in any industry standards bodies, regulatory working groups, or cross-organization consortia that are shaping the governance frameworks your sector will operate under?
  • What is your roadmap for multi-modal agentic AI capabilities, and have you identified the workflows in your organization where vision, audio, or interface-interaction capabilities would unlock new value?
  • How are you preparing your workforce — not just technically, but organizationally and culturally — for the blended human-agent model that will define enterprise operations over the next five years?
  • What is your organization’s long-term vision for the role of human professionals alongside AI agents, and have you articulated that vision explicitly enough for employees to understand how their roles will evolve rather than disappear?